UNLABELLED Quantitative assessment of (11)C-erlotinib uptake may be useful in selecting non-small cell lung cancer (NSCLC) patients for erlotinib therapy. The purpose of this study was to find the optimal pharmacokinetic model for quantification of uptake and to evaluate various simplified methods for routine analysis of (11)C-erlotinib uptake in NSCLC patients. METHODS Dynamic (15)O-H2O and (11)C-erlotinib scans were obtained in 17 NSCLC patients, 8 with and 9 without an activating epidermal growth factor receptor mutation (exon 19 deletion or exon 21-point mutation). Ten of these subjects also underwent a retest scan on the same day. (11)C-erlotinib data were analyzed using single-tissue and 2-tissue-irreversible and -reversible (2T4k) plasma input models. In addition, several advanced models that account for uptake of radiolabeled metabolites were evaluated, including a variation of the 2T4k model without correcting for metabolite fractions in plasma (2T4k-WP). Finally, simplified methods were evaluated-that is, SUVs and tumor-to-blood ratios (TBR)-for several scan intervals. RESULTS Tumor kinetics were best described using the 2T4k-WP model yielding optimal fits to the data (Akaike preference, 43.6%), acceptable test-retest variability (12%), no dependence on perfusion changes, and the expected clinical group separation (P <0.016). Volume of distribution estimated using 2T4k-WP and 2T4k were highly correlated (R(2) = 0.94). Similar test-retest variabilities and clinical group separations were found. The 2T4k model did not perform better than an uncorrected model (2T4k-WP), probably because of uncertainty in the estimation of true metabolite fractions. Investigation of simplified approaches showed that SUV curves normalized to patient weight, and injected tracer dose did not reach equilibrium within the time of the scan. In contrast, TBR normalized to whole blood (TBR-WB) appeared to be a useful outcome measure for quantitative assessment of (11)C-erlotinib scans acquired 40-60 min after injection. CONCLUSION The optimal model for quantitative assessment of (11)C-erlotinib uptake in NSCLC was the 2T4k-WB model. The preferred simplified method was TBR-WB (40-60 min after injection) normalized using several whole-blood samples.